Carrying Out a Chi-Square Test for Homogeneity or Independence

Introduction

In AP Statistics, chi-square tests help us study categorical data by comparing what we actually observe to what we would expect if there were no real relationship or no real difference. In this lesson, students, you will learn how to carry out a chi-square test for homogeneity or independence and how to tell when each one is the right choice. These tests are common in real life 🧠, from comparing how different groups vote to checking whether two variables are related.

Lesson objectives

By the end of this lesson, students, you should be able to:

Explain the key ideas and vocabulary behind chi-square tests for homogeneity and independence.
Choose the correct chi-square procedure for a categorical data situation.
Compute expected counts and the chi-square test statistic using the formula $\chi^2 = \sum \frac{(O-E)^2}{E}$.
Use the proper degrees of freedom, $df = (r-1)(c-1)$.
Interpret the result in context using a conclusion about association, homogeneity, or independence.

A chi-square test is not about averages or numerical measurements. Instead, it asks whether the counts in categories differ more than we would expect by random chance. That makes it a powerful tool for comparing groups and spotting relationships in real-world data 📊.

What Homogeneity and Independence Mean

The two tests in this lesson are very closely related, but they answer slightly different questions.

A chi-square test for homogeneity is used when we want to compare the distribution of one categorical variable across several different populations or groups. For example, students, suppose a researcher wants to know whether three different schools have the same distribution of student club preferences. Each school is a separate population, and the question is whether the category proportions are the same across all of them.

A chi-square test for independence is used when we take one random sample from a single population and measure two categorical variables for each individual. The question is whether the variables are related. For example, a survey might record both a student’s grade level and whether they prefer online or in-person homework help. The test checks whether the two variables are independent or associated.

The math for both tests is the same. The difference is in the wording of the problem:

Use homogeneity when comparing several populations or treatments.
Use independence when studying two categorical variables in one random sample.

In both cases, we organize the data in a two-way table and compare observed counts to expected counts.

Step 1: Check the Conditions

Before calculating anything, students, always verify the conditions. AP Statistics expects you to show that the chi-square procedure is appropriate.

1. Randomness

The data should come from a random sample or random assignment.

For a test of independence, the sample should be random from the population.
For a test of homogeneity, each group or population should be represented by a random sample or random assignment.

2. Independence of observations

Individual observations should be independent. If sampling without replacement, the sample size should be less than $10\%$ of the population when applicable. This is the $10\%$ condition.

3. Large expected counts

All expected counts should be at least $5$.

Expected counts are found using

$$E = \frac{(\text{row total})(\text{column total})}{\text{grand total}}.$$

If any expected count is less than $5$, the chi-square approximation may not be valid for AP Statistics.

These conditions matter because the chi-square distribution is an approximation. If the data are not random or expected counts are too small, the conclusion may not be trustworthy.

Step 2: State the Hypotheses

The hypotheses depend on whether you are doing homogeneity or independence.

For a test of independence:

$H_0$: The two categorical variables are independent.
$H_a$: The two categorical variables are associated.

For a test of homogeneity:

$H_0$: The distribution of the categorical variable is the same for all populations or groups.
$H_a$: At least one population or group has a different distribution.

Notice that the null hypothesis always represents “no difference” or “no relationship.” The alternative says there is a difference or association somewhere.

Example: Suppose a school wants to know whether preferred lunch type is related to grade level. The hypotheses would be:

$H_0$: Lunch preference and grade level are independent.
$H_a$: Lunch preference and grade level are associated.

Step 3: Calculate Expected Counts

Expected counts tell us what the table would look like if the null hypothesis were true. They are based on the assumption that there is no association or no difference in distributions.

For each cell in the table, use

$$E = \frac{(\text{row total})(\text{column total})}{\text{grand total}}.$$

Example

Suppose a table compares favorite type of exercise across two age groups. If the row total for “Teen” is $80$, the column total for “Yoga” is $50$, and the grand total is $200$, then the expected count for Teen and Yoga is

$$E = \frac{(80)(50)}{200} = 20.$$

That means if age and exercise preference were unrelated, we would expect about $20$ teens to choose yoga.

Expected counts are important because they give the “fair share” under the null model. The chi-square test measures how far the observed counts are from these expected counts.

Step 4: Compute the Chi-Square Test Statistic

Once you have observed counts $O$ and expected counts $E$, compute the chi-square statistic:

$$\chi^2 = \sum \frac{(O-E)^2}{E}.$$

This formula adds up the contribution from every cell in the table. Each term is always nonnegative because it is squared.

How to read the statistic

If observed counts are close to expected counts, $\chi^2$ will be small.
If observed counts are far from expected counts, $\chi^2$ will be large.

A large $\chi^2$ value suggests that the null hypothesis may not fit the data well.

Mini example

Imagine a cell has $O=30$ and $E=20$.

Then the contribution is

$$\frac{(30-20)^2}{20} = \frac{100}{20} = 5.$$

If another cell has $O=18$ and $E=20$, the contribution is

$$\frac{(18-20)^2}{20} = \frac{4}{20} = 0.2.$$

The first cell contributes much more because it differs more from expectation.

Step 5: Find the Degrees of Freedom

For both chi-square homogeneity and independence tests, the degrees of freedom are

$$df = (r-1)(c-1),$$

where $r$ is the number of rows and $c$ is the number of columns.

For example, if a table has $3$ rows and $4$ columns, then

$$df = (3-1)(4-1) = 2 \cdot 3 = 6.$$

Degrees of freedom matter because the chi-square distribution depends on the table size. Bigger tables have more degrees of freedom, which changes the shape of the reference distribution.

Step 6: Find the $p$-Value

After calculating $\chi^2$ and $df$, use the chi-square distribution to find the $p$-value.

The $p$-value is the probability of getting a chi-square statistic at least as large as the one observed, assuming the null hypothesis is true. In symbols, this is the right-tail area beyond the test statistic.

A small $p$-value means the observed table would be unusual if the null were true. A large $p$-value means the data are not unusual enough to reject the null.

On the AP exam, you may use a calculator or software to find the $p$-value. You should still know what the value means in context.

Step 7: Make a Conclusion in Context

Your final conclusion should always be written in context, students. Do not just say “reject the null.” Say what that means for the situation.

If the $p$-value is less than the significance level $\alpha$, then reject $H_0$.

For independence, conclude that there is convincing evidence of an association between the two categorical variables.
For homogeneity, conclude that there is convincing evidence that the distribution of the categorical variable is not the same across all groups.

If the $p$-value is greater than $\alpha$, then fail to reject $H_0$.

For independence, there is not convincing evidence of an association.
For homogeneity, there is not convincing evidence that the distributions differ.

Remember, failing to reject $H_0$ does not prove that $H_0$ is true. It only means the sample does not provide strong enough evidence against it.

How to Choose the Right Chi-Square Procedure

Choosing the right test is a major AP Statistics skill. Here is a simple way to decide:

If you have one sample and two categorical variables, use a chi-square test for independence.
If you have two or more groups or populations and want to compare the distribution of one categorical variable, use a chi-square test for homogeneity.

A helpful clue is the wording:

“Is there a relationship between…” often points to independence.
“Do the distributions differ across groups?” often points to homogeneity.

Both tests use the same formula, the same expected count rule, and the same conclusion structure. The biggest difference is the research setting.

Conclusion

Chi-square tests for homogeneity and independence are key tools for analyzing categorical data in AP Statistics. They help answer important questions like whether groups differ or whether two variables are related. The process is systematic: check conditions, state hypotheses, compute expected counts, calculate $\chi^2$, find the degrees of freedom, determine the $p$-value, and write a conclusion in context.

students, the most important idea to remember is that these tests compare observed counts to expected counts under a null model. If the differences are too large to be explained by chance alone, the data provide evidence against the null hypothesis. That is the heart of chi-square inference ✨.

Study Notes

Chi-square tests are used for categorical data and compare observed counts to expected counts.
Use a chi-square test for independence when studying two categorical variables in one random sample.
Use a chi-square test for homogeneity when comparing the distribution of one categorical variable across multiple groups or populations.
The hypotheses for independence are $H_0$: variables are independent and $H_a$: variables are associated.
The hypotheses for homogeneity are $H_0$: all group distributions are the same and $H_a$: at least one distribution differs.
Expected counts are found with $E = \frac{(\text{row total})(\text{column total})}{\text{grand total}}$.
Every expected count should be at least $5$ for the chi-square test to be valid in AP Statistics.
The chi-square statistic is $\chi^2 = \sum \frac{(O-E)^2}{E}$.
Degrees of freedom are $df = (r-1)(c-1)$.
A large $\chi^2$ value means observed counts are far from expected counts.
A small $p$-value gives evidence against $H_0$.
Conclusions must be written in context and should mention association or differences in distributions.
Failing to reject $H_0$ does not prove $H_0$ is true.
Both tests use the same calculation process; the main difference is the study design and wording of the problem.